Why BD failed? AMD Ex-Employee speaks out!

[CDdude55]

Originally Posted by Horrux

History of microprocessors must be your weak point. Intel bribed OEMs and distributors to the tune of billions so that they would exclude AMD, Cyrix and other (now defunct) makers of x86 CPUs, over a number of years. How does that compute in your previous post??

Already said that was true. I'm not an Intel fanboy just to let you know, my system is all AMD.

History also shows that AMD owns a lot to Intel for copying their designs up until the 90's. They needed Intel.

But the bigger the corporation the bigger the frauds.

[Benetanegia]

Originally Posted by Super XP

I read that, the 8150 blows away the x6 1100T out of the water in that review, but its still dated at Oct 13, 2011. So what original review was done on Oct 12, 2011, so it took them one day for the updates?

I'd like to know what "blow out of the water" means to you. I've not seen a single app in that review where the 8150 is even 15% faster than 1100T and it loses in many tests. Most of the times they are about equal.

[erocker]

I think some folks need to realize that the FX series consists of more than just an 8 core CPU.

Here's a FX 4 core vs. a PII 955: http://www.xtremesystems.org/forums/...le-of-the-X4’s

Any way you look at it, it's a failure here.

[Dent1]

Originally Posted by Benetanegia

I'd like to know what "blow out of the water" means to you. I've not seen a single app in that review where the 8150 is even 15% faster than 1100T and it loses in many tests. Most of the times they are about equal.

I agree the 8150 doesnt blow the 1100T out the water, but neither does the i5 and i7.

To be honest the 1100T does a good job at beating out the i5 2500K and i7 2600K in a few tests too, so get off your high horse about the AMD 8150 losing to the 1100T the odd time.

http://www.guru3d.com/article/amd-fx...mance-review/1

[LordJummy]

Originally Posted by Dent1

I agree the 8150 doesnt blow the 1100T out the water, but neither does the i5 and i7.

To be honest the 1100T does a good job at beating out the i5 2500K and i7 2600K in a few tests too, so get off your high horse about the AMD 8150 losing to the 1100T the odd time.

http://www.guru3d.com/article/amd-fx...mance-review/1

Actually the 2500K and 2600K really do blow the 1100T out of the water. What does that have anything to do with this thread though?

Deja vu vu vu vu vu

[Benetanegia]

Originally Posted by Dent1

I agree the 8150 doesnt blow the 1100T out the water, but neither does the i5 and i7.

To be honest the 1100T does a good job at beating out the i5 2500K and i7 2600K in a few tests too, so get off your high horse about the AMD 8150 losing to the 1100T the odd time.

http://www.guru3d.com/article/amd-fx...mance-review/1

I fail to see how the 1100T beating out SB in a few tests makes Bulldozer less of a failure??

SB vs. Thuban, that's two ~900 billion transistor CPU battling each other.

Bulldozer is a bloody 2 billion transistor CPU!! When are any of you going to open the eyes and see the failure that that is? And the blatant lie? 25% more transistors for doubling the cores and 80% performance increase? If Deneb had 750 million + 25% that makes 1 billion, not 2 billion unless I'm using math from this galaxy and I should be using math from Andromeda or something. It get worse if I do the math using Thuban cores, 900/6= 150, 150*4= 600, +25% =750. IF BD was a 750 million transistor CPU no one would be complaining, but it is again, 2 bloody billion transistors of pure fail.

2000 million / 150 million = 13.33, hello Phenom II X12??

[Super XP]

Originally Posted by Benetanegia

I'd like to know what "blow out of the water" means to you. I've not seen a single app in that review where the 8150 is even 15% faster than 1100T and it loses in many tests. Most of the times they are about equal.

"Blow out of the water" to me means about 5% to 10%. Perhaps Blow out of the water is a little strong, how about FX 8150 blows smoke in the face of the PII x6 1100T

On a side note, screw how many transistors Bulldozer has, that is a none issue. For being quite new in design, and nothing we've seen before, it does O.K. we only need to now hope AMD learns from it and releases a better version aka Piledriver.

Anyhow in regards to that 10% performance increase for Piledriver over Bulldozer, I now realize, that is full of shit. That so called AMD slide was created at the same time Bulldozer's slide was created. Those slides were all based on Bulldozer's estimated speeds via paper and/or the ES stepping. It was after the B0 stepping when AMD must have realized they were in deep shit. So right now it's all speculation as to how much juice AMD can squeeze out of Bulldozer to justify calling it Piledriver. I think they can squeeze out a lot of power out of Bulldozer, they just have to iron out some stuff first.

I don't know right now, but I plan on finding out with my old ATI contacs.

[Neuromancer]

Originally Posted by Benetanegia

If Deneb had 750 million + 25% that makes 1 billion, not 2 billion unless I'm using math from this galaxy and I should be using math from Andromeda or something.

Actually not sure what math you are using. 750 million + 33% would be 1 billion. + 25% is less than 940 million

Originally Posted by Super XP

"Blow out of the water" to me means about 5% to 10%. Perhaps Blow out of the water is a little strong, how about FX 8150 blows smoke in the face of the PII x6 1100T

Please do not mention smoke around the FX-8150 crowd lol

How about non-sensationalist descriptors, like improved or increased, not dominates, slaughters, murders, kills, slaps it around and calls it betty. etc

It is a good idea to use normal descriptors as it prevents the rabid posters from getting bent out of shape (Something I need to work on myself, not calling you out )

[Benetanegia]

Originally Posted by Super XP

On a side note, screw how many transistors Bulldozer has, that is a none issue. For being quite new in design, and nothing we've seen before, it does O.K. we only need to now hope AMD learns from it and releases a better version aka Piledriver.

Of course it is an issue. A big one. There's no way they can ever dream of really competing in price. It also means that they just cannot add more logic in order to fix things unless they really do fix things. No "trying" allowed, only success is allowed.

And new design or not, the design was suposed to reduce the transistors required per-core or per-performance unit, but it has made it skyrocket instead. Such a huge difference from what it was claimed to be (+25% aka 750 million) to what it really is, just cannot come from bad execution alone. The concept is flawed from the beginning.

[Dent1]

Originally Posted by Benetanegia

I fail to see how the 1100T beating out SB in a few tests makes Bulldozer less of a failure??

SB vs. Thuban, that's two ~900 billion transistor CPU battling each other.

Bulldozer is a bloody 2 billion transistor CPU!! When are any of you going to open the eyes and see the failure that that is? And the blatant lie? 25% more transistors for doubling the cores and 80% performance increase? If Deneb had 750 million + 25% that makes 1 billion, not 2 billion unless I'm using math from this galaxy and I should be using math from Andromeda or something. It get worse if I do the math using Thuban cores, 900/6= 150, 150*4= 600, +25% =750. IF BD was a 750 million transistor CPU no one would be complaining, but it is again, 2 bloody billion transistors of pure fail.

2000 million / 150 million = 13.33, hello Phenom II X12??

I'm not implying that bulldozer is any less of a failure. I'm saying that your argument isn’t logical or objective if you're have one rule for the SB and another rule for the Bulldozer when the 1100T outruns both in the odd test.

Seems convenient that when I point out the 1100T beating out the 2500/2600k you start changing the discussion or flipping the script to transistor counts to justify your argument. I'm not talking about transistors here I’m talking objective performance based on Guru of 3D results regardless of the manufacturing process.

[Benetanegia]

Originally Posted by Dent1

I'm not implying that bulldozer is any less of a failure. I'm saying that your argument isn’t logical or objective if you're have one rule for the SB and another rule for the Bulldozer when the 1100T outruns both in the odd test.

Seems convenient that when I point out the 1100T beating out the 2500/2600k you start changing the discussion or flipping the script to transistor counts to justify your argument. I'm not talking about transistors here I’m talking objective performance based on Guru of 3D results regardless of the manufacturing process.

Now you are grasping at straws arent' you? The X6 is a native 6 core CPU, top of the AMD line when released, SB is a 4 core mainstream CPU, even though everyone sees convenient to forget that. BD is supposed to be an 8 core CPU and it sure is a top of the line CPU with its 2 billion transistors unless AMD pretends to release a 4 billion one. A 6 core CPU beating a 4 core CPU in multi-threaded apps is not news, it should happen 100% of the times, a 900 million transistor 6 core CPU (or even a 4 core CPU) beating a 2 f billion transistor CPU IS news, bad news, very very bad news, it'0s a pathetic joke. Usually the tests where 1100T is faster than SB and BD the 980X from Intel smokes em all.

(and I did say smoke to piss neuromancer off )

[Crap Daddy]

Originally Posted by erocker

I think some folks need to realize that the FX series consists of more than just an 8 core CPU.

Here's a FX 4 core vs. a PII 955: http://www.xtremesystems.org/forums/...le-of-the-X4’s

Any way you look at it, it's a failure here.

I've seen that and it's bad. I wondered from the start, apart from the dissapointment of the top dog, the 8150, what will AMD do with the lower end. I can't see how they can sell this line of FX if there's still Phenom stock.

[Horrux]

Originally Posted by CDdude55

Already said that was true. I'm not an Intel fanboy just to let you know, my system is all AMD.

History also shows that AMD owns a lot to Intel for copying their designs up until the 90's. They needed Intel.

But the bigger the corporation the bigger the frauds.

My apologies, I posted before being finished with all the existing posts.

[Dent1]

Originally Posted by Benetanegia

Now you are grasping at straws arent' you? The X6 is a native 6 core CPU, top of the AMD line when released, SB is a 4 core mainstream CPU, even though everyone sees convenient to forget that. BD is supposed to be an 8 core CPU and it sure is a top of the line CPU with its 2 billion transistors unless AMD pretends to release a 4 billion one. A 6 core CPU beating a 4 core CPU in multi-threaded apps is not news, it should happen 100% of the times, a 900 million transistor 6 core CPU (or even a 4 core CPU) beating a 2 f billion transistor CPU IS news, bad news, very very bad news, it'0s a pathetic joke

(and I did say smoke to piss neuromancer off )

Again with the transistor talk when we are talking about performance based on the results of Guru of 3D, not performance based on the metallic apparatus used.
http://www.guru3d.com/article/amd-fx...mance-review/1

Originally Posted by Benetanegia

Usually the tests where 1100T is faster than SB and BD the 980X from Intel smokes em all.

The 980X should smoke them all since it bears a $800-100 price tag.

But price and transistor count aside, the 1100T beats out the 2500K/2600K in certain tests. Hence the 2500K/2600K should also be criticised for being unable to outrun the K10 as much as the Bulldozer are being criticised for being unable to do so.

[Crap Daddy]

I can link you to this comaprison from anand. If I'm not mistaken the 1100 beats the 2600 only in power consumption:

http://www.anandtech.com/bench/Product/203?vs=287

And the transistor count comparison counts also the IGP on Sandy?

[Benetanegia]

Originally Posted by Dent1

Again with the transistor talk when we are talking about performance based on the results of Guru of 3D, not performance based on the matalic aparatus used.



The 980X should smoke them all since it bears a $800-100 pricetag.

But price and transistor count aside, the 1100T beats out the 2500K/2600K in certain tests hence any negative it's only right the 2500K/2600K are criticised for being unable to outrun the K10 as much as the Bulldozer are being criticised for being unable to do so.

Sorry but no. Sandy was never designed to win those benches. It's a mainstream CPU where the effort was put into making single threaded or light threaded apps runs faster and damn sure it does, smoking both BD and 1100T on those tasks.

Now, both Thuban and BD were designed for multi-threaded apps, 1100T succeeds, bravo, but bulldozer is fail. Both in multi-threaded (many not all) and single threaded tasks.

980X should not smoke them all, price tag is irrelevant and only a consequence of lack of competition. It's a 1.1 billion transistor chip, so much smaller than Bulldozer and a lot cheaper to produce. Don't blame Intel for charging what they do, blame capitalism if that makes you feel better, or pray to some saints for some competition in the future. Either way Bulldozer sucks.

[Crap Daddy]

The 980X is discontinued as far as I know, so it's over. SB-E wil have, starting next month a six-core CPU i7-3930K rumoured to be around 500$-600$. Now that's "smoking"

[LordJummy]

Originally Posted by Dent1

But price and transistor count aside, the 1100T beats out the 2500K/2600K in certain tests. Hence the 2500K/2600K should also be criticised for being unable to outrun the K10 as much as the Bulldozer are being criticised for being unable to do so

Your posts are highly illogical.

First off you are way off topic, and so is the guy you are arguing with.

Secondly, the 2x00K SB chips should not be criticized in the same fashion BD is. They far surpassed their expectations as mainstream chip(s). In chips of this fashion compromises must be made. The 6 core 1100T should be expected to win a few multi threaded tasks that the SB chips are not suited for.

You seem to be just making any excuse possible at this point to justify your arguments. Perhaps you should try opening your mind a bit and letting go of your biased mentality. You will then be able to see BD for what it is, from an honest perspective

Just my opinions though, take them however you'd like.

[Benetanegia]

Originally Posted by Crap Daddy

The 980X is discontinued as far as I know, so it's over. SB-E wil have, starting next month a six-core CPU i7-3930K rumoured to be around 500$-600$. Now that's "smoking"

Yeah I wonder if they will still defend BD when SB-E smokes it in EVERY posible way...
Because as I'm seing right now, these people are deluded in thinking that BD is and always has been competition for SB. As if transistor budget didn't matter or wasn't a clear (as water) indicative of what market segment the resulting chip was suposed to belong to.

In retrospective I don't know why Fermi suffered such a bad reputation, I mean it smoked the HD5770 right?

[Crap Daddy]

No. Fermi at least held the crown of the fastest GPU on the planet although huge, hot and power hungry-

[Dent1]

Originally Posted by Benetanegia

Yeah I wonder if they will still defend BD when SB-E smokes it in EVERY posible way...
Because as I'm seing right now, these people are deluded in thinking that BD is and always has been competition

I see very few people defending Bulldozer. Me myself is defending the Phenom II X6 1100T which appears to be holding its own against Bulldozer and SB at times. Fvck Bulldozer.

[fullinfusion]

God I feel like im in a room full of toddlers! mines bigger then yours, this is that!!! really who cares? Anyways let AMD iron out there flaws and lets close these usless BD threads and actually make some useful HELPFULL posts!

I cant believe I allow my mail server to update me on such childishness commented threads! *unsubscribed*

Oh and to the upper intel fan kids on this page... the 1100T and 1090T does give the 25-26**K's a good run for its money in some marks.. Some of ya need to read and view the comparisons many of us have posted to compare such benchmarks and the actual score..

Thats all

[D4S4]

Originally Posted by Benetanegia

A 6 core CPU beating a 4 core CPU in multi-threaded apps is not news, it should happen 100% of the times, a 900 million transistor 6 core CPU (or even a 4 core CPU) beating a 2 f billion transistor CPU IS news, bad news, very very bad news, it'0s a pathetic joke.

then what would you call intel's itanium fiasco? it's a fucking ginormous chunk of silicon yet it sucks. true that it's based on a different architecture but like bulldozer, the software simply isn't optimized for it. ain't no way in hell those 2 billion transistors are there just to generate heat.

[Benetanegia]

Originally Posted by D4S4

then what would you call intel's itanium fiasco? it's a fucking ginormous chunk of silicon yet it sucks. true that it's based on a different architecture but like bulldozer, the software simply isn't optimized for it. ain't no way in hell those 2 billion transistors are there just to generate heat.

Didn't you already said it all in 2 words? Itanium (1) fiasco (2).

What else is there to say? Previous and future failures do not mitigate the failure that Bulldozer is right now plain and simple.

[qubit]

Originally Posted by D4S4

then what would you call intel's itanium fiasco? it's a fucking ginormous chunk of silicon yet it sucks. true that it's based on a different architecture but like bulldozer, the software simply isn't optimized for it. ain't no way in hell those 2 billion transistors are there just to generate heat.

Unfortunately, it looks like they are. Take a look at these two articles I posted about some way back in this thread.

http://www.techpowerup.com/forums/sh...34#post2430034

AMD royally screwed up with this launch and their future as an x86 CPU manufacturer doesn't look too rosy right now.